1279 gossiplb inform and decide have bugs #1348
Merged
+730
−133
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
nlslatt
requested review from
lifflander,
cz4rs,
JacobDomagala,
jstrzebonski,
nmm0,
PhilMiller and
ppebay
lifflander
approved these changes
Apr 5, 2021
Closed
PhilMiller
reviewed
Apr 7, 2021
PhilMiller
reviewed
Apr 7, 2021
PhilMiller
reviewed
Apr 7, 2021
| int n_transfers_ = 0; | ||
| }; | ||
|
|
||
| struct GossipRejectionStatsMsg : collective::ReduceTMsg<RejectionStats> { |
There was a problem hiding this comment.
@lifflander Would it be possible to avoid the need to define types like this, by just passing through constructor arguments to the encapsulated type T in ReduceTMsg<T>?
PhilMiller
reviewed
Apr 7, 2021
nlslatt
force-pushed
the
1279-gossiplb-inform-and-decide-have-bugs
branch
from
c8b5260 to
242282c
Compare
Codecov Report
@@ Coverage Diff @@
## develop #1348 +/- ##
===========================================
+ Coverage 82.50% 82.60% +0.09%
===========================================
Files 759 759
Lines 28279 28506 +227
===========================================
+ Hits 23333 23546 +213
- Misses 4946 4960 +14
|
7 tasks
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Includes many bug fixes and improvements to GossipLB. The following LB args were added:
trials(default3): how many times to repeat the requested number of iterations, hoping to find a better imbalance; helps if it’s easy to get stuck in a local minimumdeterministic(default false): for debugging purposes, make the migration decision deterministic assuming deterministic loads (for testing on deterministic loads, consider using the driver in development under Implement general test driver for LB testing #1265)inform(default SyncInform): choice of gossiping approach0): synchronous propagates after all recvs for a round, but has sync cost (matches LBAF approach)1): asynchronous propagates after the first recv for a round, but avoids sync costordering(default Marginal): order in which to evaluate local objects for migration0): use the unordered_map iteration order1): order by ascending element ID2): order by descending load starting with the object of marginal load, then order ascending for larger loadscmf(default NormByMax): the algorithm used for computing the CMF0): remove processors from the CMF as soon as they exceed the target (e.g., processor-average) load; use a CMF factor of 1.0/x, where x is the target load1): do not remove processors from the CMF that exceed the target load until the next iteration; use a CMF factor of 1.0x, where x is the maximum of the target load and the most loaded processor in the CMF2): do not remove processors from the CMF that exceed the target load until the next iteration; use a CMF factor of 1.0x, where x is the load of the processor that is computing the CMFrollback(default true): whether to roll back to an earlier iteration if it had the best imbalancetargetpole(default false): whether to replace the processor-average load with the max of that and the maximum object load, effectively redefining overloaded/underloaded based on the longest pole load when it exceeds the processor-average loadNote that I have a both a develop-based branch (this PR;
1279-gossiplb-inform-and-decide-have-bugs) and a release-based branch (1279-gossiplb-inform-and-decide-have-bugs-release).Closes #1279